组合NNSC收缩技术和改进四阶PDE的MMW图像恢复

MMW image restoration based on the combination of NNSC shrinkage technique and modified four-order PDE model

毫米波(MMW)图像含有大量未知噪声,仅用一种方法恢复的效果较差,因此结合非负稀疏编码(NNSC)收缩技术和改进四阶偏微分方程(PDE)模型的优点,提出了一种基于组合处理的MMW图像恢复方法。NNSC收缩法具有自适应消噪图像的特性,和数据的属性无关;而改进四阶PDE能够消除二阶PDE产生的阶梯效应,同时避免光滑区域不平整的现象,具有较好的图像恢复效果。分别采用模拟的和真实的MMW图像进行测试,并用相对信噪比(RSNR)进行评判,实验表明,与NNSC收缩、基于四阶的PDE模型以及小波收缩等方法相比,所提出的方法能够有效地用于MMW图像的恢复。

A millimeter wave （MMW） image contains much unknown noise, and the restoration quality is worse only using sole method. Combined the advantages of non - negative sparse coding （ NNSC）shrinkage technique and modified four - order partial differential equation （PDE）, a combination based MMW image restoration method is proposed. The NNSC shrinkage method behaves adaptive property of denoising and is irrelative to the data attribute. The modified four - order PDE can delete the stair - casing effect, generated by the two - order PDE, and avoid the un - smoothing phenomenon of flat region, and behaves the better effect on image restoration. Utilizing respectively the simulated and true MMW image to test and the relative signal noise ratio （RSNR） to measure the quality of restored results, and compared with methods of NNSC shrinkage, fourorder PDE and wavelet shrinkage, experimental results show that the method proposed here is efficient in the MMW image restoration.